Monday, April 25, 2016

The woolly bear presidential election outlook, 2016

In the age of cell phones, accurate polling of the electorate has become difficult. In a world where a disproportionate percentage of people answering landlines for pollsters is white and over 50, we desperately need a new method of predicting elections. As the 2016 presidential election looms, a crack team of UC Davis innovators has a promising new source of information, woolly bear caterpillars (Platyprepia virginalis).

A woolly bear contemplates the madness of the 2016 election cycle while resting on its preferred host plant, a coastal lupine. Photo: Eric LoPresti
Rick Karban, a UC Davis professor of entomology, has tracked woolly bear caterpillar abundance since the early 1980’s at Bodega Bay, California. Each March, Karban censuses the same patches of lupine that he has for over 30 years. The study asks a vexing question: Why are there are so many caterpillars in some years and so few in others? Many insects, including pests cycle like this, therefore it is of keen interest to many. Dozens of papers later, Karban, his students, and his collaborators have answered a great many questions, including how caterpillars deal with parasites, whether population cycles are influenced by rain, whether caterpillars enjoy eating plant hairs, and how caterpillars avoid their predators.

The population highs and lows seem random at a first pass, a jagged line moving up and down each year. 

The collected data, 1983-2015, full data available here

What separates high years from low years? These motivated researchers have found a striking pattern in this data. This data set includes eight presidential election years, with four Democrat and four Republican victories. Plotted with colors corresponding to the party association of the winner, the pattern becomes obvious.

Red corresponds to Republican presidential victories and blue to Democrats. To reiterate: this is actual data!
Woolly bears have years of high abundance when Democrats win and low when Republicans win. The average woolly bear abundance was 0.21 (+ 0.07 se) woolly bears per lupine in Republican years and 1.96 (+0.27 se) in Democratic years.  This data shows that woolly bear abundance in March is a good predictor of presidential victories in the general election.

It is tempting to assume that woolly bears are Democrats (and were particularly thrilled by second-term Bill Clinton), but we cannot exclude the possibility that their abundance is a protest gesture. 

Note that 2016 is not included on the preceding two graphs. For about a year, news sources have made predictions about the primary race and have even speculated about the general election. Given their wildly erroneous predictions thus far for both primaries, trusting their predictions for the general election seems ill-advised. The woolly bears, on the other hand, have a 100% accurate prediction record over the past 30 years. In years of low abundance, a Republican is elected, and in years of high abundance, a Democrat. 

Therefore, we are pleased to announce the woolly bears’ prediction. In mid-March of this year, Karban censused the woolly bears for their opinion on this volatile election year where no subject seems out of bounds and the populist wings of each party have come out like no election in recent memory. Even the woolly bears seem hesitant this year. 

Full data, including this year's census (conducted in March).
A superficial examination suggests that 2016 will be a Republican year – woolly bear abundance is not particularly high. However, looking a little closer, it may not be. The number of woolly bears per lupine bush in 2016 (0.53) is higher than the average Republican year by 152% and is 36% above the highest Republican year ever recorded (1988). However, it is only 27% of an average Democratic year and still only 36% of the lowest Democratic year (2008). This result is without presidential precedent in the last 30 years.

We suspect that the Republicans have the edge. However, a valid hypothesis would be a third-party winner, such as a right-leaning independent (a logical placeholder in between Democrats and Republicans). Perhaps Donald Trump will take particular interest in our data. Alternately, a contested Republican convention could produce a fractured party and the old Republican woolly bear average would not accurately represent the mean caterpillar abundances seen by this new party.

The mainstream media have been shockingly inaccurate in their predictions so far, even despite complex and supposedly accurate statistical models. We need a new strategy to predict key events such as the 2016 presidential election. Rather than trust the opinion of a few people with a pulpit, the historically robust predictions of this population of caterpillars may serve as a better guide.

A congregation of woolly bears meets on a lupine flower spike, presumably to discuss politics. Photo: Eric LoPresti

(This research has been in progress and was presented at ESA 2014)

This post was written by Eric LoPresti, Mikaela Huntzinger, Patrick Grof-Tisza, Ian Pearse, and, of course, Rick Karban (who we suspect is not fooling these infallible caterpillars with his Bernie Sanders impersonation). 

Rick Karban/Bernie Sanders. Who is who? Photo on left stolen from, right: Mikaela Huntzinger.


  1. I would ask why you believe that the caterpillars are determining the outcome of these elections. Why is it not that the elections are determining the change in caterpillar populations. Correlation does not mean causality, I learned in statistics class.

  2. Just look at them. Do you think they'd sit idly by, rather than rally (or protest - we don't know)?

  3. Did anyone else notice how the caterpillar in first photo resembled Trump? Yellow-gold hair on top of head- while rest of body hair is gray?